Abstract
Measurements of the isotope effect for impurity diffusion of iron in niobium and of the enhancement of niobium self-diffusion by iron addition, taken with the previous measurements of self-diffusion coefficients for niobium and impurity diffusion coefficients for iron in niobium, allow us to demonstrate that neither a short- nor a long-range vacancy model gives a satisfactory explanation of the fast diffusion of iron in niobium. It is also shown that diffusion models based on linear or plane defects, interstitials, simple interchange or ring mechanisms are not in agreement with experimental results.
By analogy with the mechanism suggested by Miller (1969 a, b, c) and Miller and Edelstein (1969) for high diffusivities in lead, a dissociative mechanism, involving an equilibrium between substitutional iron atoms and interstitial iron atoms, is judged acceptable. Iron diffusion is then effected by vacancy-interstitial pairs. This mechanism, which gives a completely satisfactory explanation of experimental results, is shown to be responsible also for the abnormally high impurity diffusivities for iron, cobalt and nickel in γ uranium, β titanium and β zirconium.